ABSTRACT
Whereas many people supposed that our only object was to produce the winning team, in reality we were equally interested in getting right the approximate rank order of all the teams. As a task the former would have been much easier. Other teams could have been starved of the most useful people. Instead we had the more difficult task of making fine discriminations between teams. The team that we forecast as coming second had to be ‘quite a good team’ but not so good as the team forecast as coming first.
